Line plot of timeseries with year as a factor

I have a dataset that you can download from

https://www.dropbox.com/s/j4qnib9fhgrluln/RadData_Hourly_2005_2020.csv?dl=0

Containing hourly measurements from a location.

For each variable, I would like to plot the measurements on a line plot with the x-axes having the day and month and multiple lines, one for each year.

I prepared the data frame with the code:

histdata_shinyApp <- read_csv("RadData_Hourly_2005_2020.csv", 
                                   header=TRUE,
                                   skip= 8,
                                   dtoken = refreshable_token)%>%
  mutate(TIMESTAMP=as.POSIXct(time, format='%Y%m%d:%H%M'),
         date = ymd(as.Date(TIMESTAMP)),
         year = format(date, "%Y"),
         month = format(date, "%b"),
         day = format(date, "%d"),
         time= hms(format(as.POSIXct(TIMESTAMP, format='%Y-%m-%d %H:%M:%S'),format = "%H:%M:%S")),
         .before=1)%>%
  mutate(month=factor(month, 
                      levels=c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"), 
                      labels=c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec")
                      ),
         Gb.i.=as.numeric(Gb.i.)
         )%>%
  pivot_longer(cols = Gb.i.:WS10m, 
               names_to = "variable")

I then tried to plot with ggplot2:

library(ggplot2)
library(dplyr)
histdata_shinyApp%>%
  filter(., variable=="Gb.i.")%>%
  ggplot() +
  geom_line(aes(x=month, y=value, group=year, color = year))+
  geom_point(aes(x=month, y=value, group=year, color = year))

But the result gives me the sequence of all the years on the x, rather than a day-month sequence and years as a factor for line color.

I obtained a good result using ggpubr. Still, I would like to do the same with ggplot.

histdata_shinyApp%>%
  filter(., variable=="Gd.i.")%>%
    ggline(x="day", y="value", add="mean_se", facet.by="month")

Solution

The key steps in this is that you're using a "summary" stat rather than an "identity" stat - i.e., ggplot should do calculations to determine point/line placements. To plot all lines across months, colouring by year use stat_summary twice:

library(tidyverse)

histdata_shinyApp <- read.csv("https://www.dropbox.com/s/j4qnib9fhgrluln/RadData_Hourly_2005_2020.csv?dl=1", 
                              header=TRUE,
                              skip= 8)%>%
  mutate(TIMESTAMP=as.POSIXct(time, format='%Y%m%d:%H%M'),
         date = ymd(as.Date(TIMESTAMP)),
         year = format(date, "%Y"),
         month = format(date, "%b"),
         day = format(date, "%d"),
         time= hms(format(as.POSIXct(TIMESTAMP, format='%Y-%m-%d %H:%M:%S'),format = "%H:%M:%S")),
         .before=1)%>%
  mutate(month=factor(month, 
                      levels=c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"), 
                      labels=c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec")
  ),
  Gb.i.=as.numeric(Gb.i.)
  )%>%
  pivot_longer(cols = Gb.i.:WS10m, 
               names_to = "variable")
#> Warning: There was 1 warning in `mutate()`.
#> ℹ In argument: `time = hms(...)`.
#> Caused by warning in `.parse_hms()`:
#> ! Some strings failed to parse, or all strings are NAs
#> Warning: There was 1 warning in `mutate()`.
#> ℹ In argument: `Gb.i. = as.numeric(Gb.i.)`.
#> Caused by warning:
#> ! NAs introduced by coercion


histdata_shinyApp%>%
  filter(., variable=="Gb.i.")%>%
  ggplot() +
  stat_summary(aes(x=month, y=value, group=year, color = year), fun = "mean", geom = "line")+
  stat_summary(aes(x=month, y=value, group=year, color = year))